Update app.py

app.py

@@ -1,16 +1,45 @@
 import gradio as gr
-from transformers import pipeline
+from langchain import HuggingFacePipeline, PromptTemplate, LLMChain
+from transformers import AutoTokenizer
+import transformers
+import torch
 
-# Load your model
-summarizer = pipeline("summarization", model="manohar02/Llama-2-7b-finetune")
+# Define the Hugging Face model
+model = "meta-llama/Llama-2-7b-chat-hf"
 
-# Define the summarization function
-def summarize(text):
-    summary = summarizer(text, max_length=2048, min_length=10, do_sample=False)
-    return summary[0]['summary_text']
+# Define the Hugging Face pipeline
+pipeline = transformers.pipeline(
+    "text-generation",  # task
+    model=model,
+    torch_dtype=torch.bfloat16,
+    trust_remote_code=True,
+    device_map="auto",
+    max_length=20000,
+    do_sample=True,
+    top_k=10,
+    num_return_sequences=1,
+    eos_token_id=AutoTokenizer.from_pretrained(model).eos_token_id
+)
 
-# Set up the Gradio interface
-iface = gr.Interface(fn=summarize, inputs="text", outputs="text", title="Text Summarizer")
+llm = HuggingFacePipeline(pipeline=pipeline, model_kwargs={'temperature': 0})
 
-if __name__ == "__main__":
-    iface.launch()
+# Define the template for summarization
+template = """
+Write a concise summary of the following text delimited by triple backquotes.
+'''{text}'''
+SUMMARY:
+"""
+
+prompt = PromptTemplate(template=template, input_variables=["text"])
+
+llm_chain = LLMChain(prompt=prompt, llm=llm)
+
+# Function to generate summary
+def generate_summary(text):
+    summary = llm_chain.run(text)
+    # Extract only the summary part from the output
+    return summary.strip()
+
+# Create a Gradio interface
+iface = gr.Interface(fn=generate_summary, inputs="text", outputs="text", title="LLaMA2 Summarizer")
+iface.launch()